BFF_gaussian_seq_update: BFF Change Point Detector Sequential Update

In elizabethriddle/BFF: Sequential Adaptive Parameter Estimation and Change Point Detection

BFF Change Point Detector Sequential Update

Description

Detects change points within data stream sequentially using a Bayesian adaptive estimation procedure and posterior p-values.

Usage

BFF_gaussian_seq_update(
  data,
  data_old,
  i,
  threshold_val = c(0.05),
  grace_period = 20,
  sw = 2000,
  alpha = 39,
  beta = 1.8,
  post_p = TRUE,
  pred_p = TRUE,
  current_mu,
  current_sigma2,
  N_values_old_old,
  D_values_old_old,
  M_values_old_old,
  N_values_old,
  D_values_old,
  M_values_old,
  mu_0_prior_value_old,
  sigma2_0_prior_value_old,
  sigma2_alpha_0_prior_value_old,
  sigma2_beta_0_prior_value_old,
  mu_0_prior_value,
  sigma2_alpha_0_prior_value,
  sigma2_beta_0_prior_value,
  anomalous_mu_threshold_sw,
  anomalous_lambda_threshold_sw,
  p_values_change_mu,
  p_values_change_lambda,
  anomalous_predictive_threshold_cal,
  p_values_predictive
)

Arguments

data	new data point
data_old	previous data point
i	data point number
threshold_val	vector of thresholds to use to assess whether new data point is a change point
grace_period	period after a change in which changes are not detected
sw	sliding window size for p-value calibration
alpha	alpha parameter for lambda beta prior
beta	beta parameter for lambda beta prior
post_p	to calculate predictive posterior p-values
current_mu	estimated mu from previous time
current_sigma2	estimated sigma2 from previous time
N_values_old_old	old old N value
D_values_old_old	old old D value
M_values_old_old	old old M value
N_values_old	old N value
D_values_old	old D value
M_values_old	old M value
mu_0_prior_value_old	old prior mean for mu
sigma2_0_prior_value_old	old prior var for mu
sigma2_alpha_0_prior_value_old	old prior alpha for sigma2
sigma2_beta_0_prior_value_old	old prior beta for sigma2
mu_0_prior_value	prior mean for mu
sigma2_alpha_0_prior_value	prior alpha for sigma2
sigma2_beta_0_prior_value	prior beta for sigma2
anomalous_mu_threshold_sw	current anomalous mu posterior p-values list
anomalous_lambda_threshold_sw	current anomalous lambda posterior p-values list
p_values_change_mu	uncalibrated mu posterior p-values
p_values_change_lambda	uncalibrated lambda posterior p-values
anomalous_predictive_threshold_cal	current anomalous predictive posterior p-values list
p_values_predictive	uncalibrated predictive posterior p-values

Value

return the change points detected by the algorithm for each of the dection procedures, for the different thresholds.

elizabethriddle/BFF documentation built on April 4, 2022, 11:51 a.m.